Residency and Fellowship Program Administrator Burnout: Measuring Its Magnitude

The concept of burnout was first presented by Freudenberger¹  and later described by Maslach and Jackson²  as a syndrome consisting of 3 elements: (1) emotional exhaustion, (2) cynicism/depersonalization (an emotional and mental separation from one's work), and (3) reduced personal accomplishment and feelings of ineffectiveness. This definition of burnout focused on individuals providing care for patients or underserved, distressed persons.²  Burnout in clinicians has been linked to a reduction in work productivity,³  poor sleep quality,⁴  high turnover,³  and an increase in alcohol consumption and drug misuse,^2,4  all factors that can negatively impact the work environment.

Survey studies have demonstrated high prevalence of burnout in US physicians and have found that burnout affects the quality of care.^5–7  The propensity for burnout to be contagious and affect other members of the health care team has also been shown.^8,9  Because they work directly with trainees and faculty, program administrators (PAs) function at the center of the graduate medical education community. To date, there has been no national study of the prevalence of burnout in PAs.

This study aimed to assess the degree of burnout and its association with consideration of leaving work among PAs from diverse US residency and fellowship programs.

Our study was open to all PAs, regardless of program accreditation status, experience level, specialty, or full-time/part-time employment status. Participants included individuals whose positions involved administration of a residency or fellowship training program. As the titles of these individuals vary across institutions, the term “program administrator” was used to describe program coordinators, residency managers, and graduate medical education fellowship coordinators, etc. Part-time and full-time statuses were based on the percentage of the participant's overall job responsibility that is exclusively dedicated to the role of a PA (part-time: ≤50%, full-time: >50%). To incentivize participation, participants external to our organization were entered into a raffle for a $100 gift card.

Solicitations were sent electronically over a 2-month period. Representatives from PA specialty associations (ie, Association of Family Medicine Administration, Association of Residency Administrators in Surgery, Coordinator Description Task Force) were contacted and asked to forward study details to their membership and specialty groups. Online PA forums, including the Association for Hospital Medical Education and LinkedIn groups, were also utilized to disseminate study information. Although the exact number of PAs nationally is unknown, it is estimated to be approximately 10 205 (Thalamus representative, written communication, April 19, 2019). Those who received the solicitation and wanted to participate were asked to e-mail the study team to be enrolled.

For this descriptive cross-sectional study, a baseline survey was distributed in July 2017 to all participants who e-mailed the study team. Survey design and setup prevented respondents from completing the survey more than once. Participation was voluntary and had no bearing on job standing. Our baseline survey included demographic questions, questions related to wellness activities, and the Copenhagen Burnout Inventory (CBI) tool.

The CBI, which has been shown to be highly correlated with the Maslach Burnout Inventory, was used to measure burnout in our study population.¹⁰  The 19-item CBI questionnaire, originally developed to study burnout in Danish human service workers (non-health care and health care workers), consists of 3 independent scales: personal, work-related, and client-related burnout.¹¹  For this study, “client” was defined as an intern, resident, fellow, or medical student.

Participants were given 1 month to complete the self-administered baseline survey. Possible scores for each subscale of the tool ranged from 0 (low) to 100 (high), with higher scores reflecting higher levels of burnout.¹¹  Some studies have created cutoffs to define higher levels of burnout (ie, score ≥50 equates to high burnout), but these figures are arbitrary.^12,13 

Each component of the CBI is assessed independently, which allows researchers to better recognize the source of burnout—personal, work related, or client related.¹¹  Notably, the full Maslach Burnout Inventory questionnaires are not available for public use unless a payment is received—a barrier for many institutions. Kristensen and colleagues¹¹  found Cronbach's alpha for internal reliability of the CBI to be high: 0.85–0.87.

Means and standard deviations were computed to ascertain personal, work-related, and client-related burnout scores across all specialties and geographic regions within our study group. A 2-sample t test was performed to test for differences in burnout scores among PAs in each specialty compared with the other studied specialties. In addition, we used a 1-way analysis of variance (unbalanced design) to test for differences in the burnout score means among geographic regions. The model adjusted for multiple comparisons using the Tukey method. Descriptive statistics for the 3 subscales of burnout, as well as all corresponding questions, were calculated using univariate analysis.

To better assess the impact of burnout on possible turnover, participants were asked, “In the last year, have you thought about resigning your position due to increased workload or work-related stressors?” We stratified by response option (yes or no) to examine how scores may relate to thoughts of resigning. We performed a 2-sample t test to measure the differences in average burnout scores between these 2 groups. The Cochran-Mantel-Haenszel test, computed to allow for stratification by confounding categorical variables, was used to identify any differences in the responses to additional questions between those who had thoughts of resigning and those who had not had thoughts of resigning. The model adjusted for age, gender, geographic location, and level of education. SAS 9.4 (SAS Institute Inc, Cary, NC) was used for all statistical analyses. Study data were collected and managed using the REDCap electronic data capture tool.¹⁴ 

The study protocol was deemed exempt by the Boston University Institutional Review Board.

Of the 1084 individuals who e-mailed the study team, 904 (83%) completed the baseline survey. We estimate that this represents 11% (1084 of 10 205) of all PAs. In addition, 29 of 42 (69%) PAs at our institution chose to participate. Two of the 933 respondents were not eligible to participate as none of their job responsibilities pertained to that of a PA, leaving 931 participants in the study group. The response rate was very high for all questions, with all or nearly all participants answering each question.

The majority of our study population was female (97%, 905 of 931) and self-identified as white or Caucasian (80%, 747 of 931). Participants came from various geographic areas' the highest percentage came from the Great Lakes region (19%, 175 of 931) followed by the Southeast (17%, 157 of 931). Nearly 60% (546 of 931) held at least a bachelor's degree and were employed in the medical education field for at least 8 years (table 1).

Table 2 provides a detailed summary of CBI score by specialty. The neurology fellowship (n = 21) and obstetrics and gynecology PAs (n = 54) had the highest personal burnout scores, with CBI scores of 66.3 and 63.3, respectively. Gastroenterology PAs (n = 23) had an average score of 42.9 for personal burnout, the lowest among all specialties studied.

Neurology PAs (n = 37) demonstrated an elevated work-related burnout score (63) relative to the other specialties. Similar to findings in the literature,¹¹  client-related burnout scores were lower than scores for the other subscales. The otolaryngology PAs (n = 19) exceeded the overall client-related burnout mean, scoring a 42.5 on average. PAs from 3 specialties had a statistically significant different work-related burnout score than PAs working in other studied specialties (gastroenterology: 42.9 versus 52.7, P = .035; ophthalmology: 61.3 versus 52.3, P = .010; and radiology: 49.6 versus 52.6, P = .018). There was also a statistically significant difference in the average personal and client-related burnout scores across geographic regions (table 3). The online supplemental material offers additional insight into participant responses to all questions within the corresponding subscales of the CBI.

Nationally, 52% (483 of 931) of participants reported having thoughts of resigning from their positions due to increased workload or work-related stressors (table 4). There was a difference in personal, work-related, and client-related burnout scores among those who had thought about resigning and those who had not contemplated resigning (table 4). After adjusting for age, gender, geographic location, and level of education, the association between having thoughts of resigning and the following 3 strata remained strong: (1) years working in the medical education field, (2) years in current position, and (3) wellness activities offered. Most respondents (91%, 704 of 777) expressed interest in attending wellness events organized by their institutions, if offered. Some respondents (14%, 124 of 902) noted that their institutions offered wellness activities specifically for PAs, which included annual retreats, workshops, counseling services, yoga/group activities, fitness challenges, and lunches.

The results of this descriptive cross-sectional study demonstrated that PAs are also at risk for burnout, with variability across specialties and geographic areas. Results showed that PAs who had contemplated resigning their positions due to increased workload or work-related stressors had higher personal, work-related, and client-related burnout scores than those who had not thoughts of resigning.

Higher CBI scores were noted among those in neurology subspecialty fellowships (66.3), obstetrics and gynecology (63.3), neurology (61.5), ophthalmology (58.7), family medicine (58.2), internal medicine (57.6), and otolaryngology (54.4). The higher scores of the aforementioned fields greatly mirror those of their specialty physician counterparts, who were similarly identified as having high rates of burnout during our study period.^15–17  Physician data reflected that high rates of burnout were seen particularly among physicians in specialties at the front line of access to care, such as practitioners in internal medicine, family medicine, emergency medicine, and neurology.^15–17 

The level of burnout in this study group is consistent with findings of other studies that have utilized the CBI to examine clinician burnout.^11,18,19  It is clear from prior studies on resident burnout that PAs share common factors, including work overload and low support at work.²⁰  Physician data reflect that burnout is a major driver of physician turnover,^21–25  and that physicians' expression of intent to leave a place of employment is correlated with actual departures.^21,26,27  While the impact of PA departure is not as well studied, and its financial impact not as well understood, the hiring, onboarding, and training of a new PA is a significant burden for training programs, and it impacts trainees and faculty.

Participation in this study was voluntary, and as such, the study population represents a sample of PAs who may or may not be representative of the entire PA population, a cohort whose exact size is unknown. In academic year 2017–2018, there were more than 11 000 ACGME-accredited residency and fellowship programs,²⁸  but with varying magnitude and structure of administrative responsibilities within these programs, the number of programs is not synonymous with the number of PAs (a PA may oversee multiple programs at his or her institution). Similarly, there are no available data reflecting the number of nonaccredited training programs or the number of PAs who support them. In addition, the total number of active members across the specialty forums we utilized is unavailable.

The findings from this study are limited by the difficulty in determining an accurate response rate given that the total number of PAs is unknown. Furthermore, it is unknown if this sample is representative of all PAs or if those with higher degrees of burnout and job dissatisfaction were more or less likely to respond to the survey. To our knowledge, this study represents the largest study of burnout among this population. Additionally, there is some validity evidence supporting the CBI in human service work but none for use in this specific population in the United States—respondents may not have interpreted questions as intended, and the 3 burnout scales may not truly measure the same constructs in PAs.

As next steps, longitudinal studies should be conducted to explore variations in burnout levels in PAs throughout the academic year, whether those considering leaving their positions actually did so, and any impact that interventions may have in mitigating this effect.

These findings suggest that burnout among a cohort of PAs from across the country is high, and those who have considered leaving their positions had higher burnout levels than those who have not.